Research Methods
- Created by: Hannah Wiersma
- Created on: 19-05-13 16:30
Laboratory Experiments
- DIrectly manipulates IV to see effect of DV. High controlled conditions.
- +: high level of control, replicable, cause and effect.
- -: artificiality makes low validity, investigator and participant effects high risk.
- Sometimes cannot give full informed consent as demand characteristics.
Field Experiments
- Directly manipulates IV to see effect on DV, in natural setting.
- Higher ecological validity, reduction in participant effects, demand characteristics. Can give cause and effect.
- Less control on extraneous variables and participant sample. Cannot replicate exact and more time consuming.
- Consent, deception and right to withdraw may occur as participants sometimes dont know they are in experiment.
Natural Experiments
- Researcher takes advantage of natural happening to see effect on DV, not manipulated.
- Useful when ethically/impossible to create experiment in lab. High ecological validity.
- Cannot control extraneous variables, low internal validity. Hard to define cause and effect, less had to control participant sample and cannot replicate or generalise as can be one off.
- Consent, right to withdraw and confidentiality, important issues.
Correlational Analysis
- Technique for analysing data, measures strength of relationship with 2 variables. Can show: Positive (both increase), negative (1 increase, 1 decreases)or no correlation.
- Can establish precise relationship, allows research in things you cannot experiment with because of ethical, practical issues.
- Cannot give casue and effect, only measures linear relationships not curvilinear.
- Consent, confidentiality, right to withdraw issues sometimes as some unaware data being used.
- Correlation coefficient: measures strength or relationship(correlation) Goes from -1 to +1.
Naturalistic Observations
- Researcher observes participants in own envioronment, no manipulation of variables deliberatley.
- High ecological validity, participants behave naturally, used for preliminary research tool suggesting hypotheses for further research.
- No control on extraneous variables.
- Privacy, confidentiality, consent important issues.
When designing you need to consider:
- Obsever bias
- Adressing ethical issues
- How are you recording data
- Identifying appropriate behavioural categories
- Do you do a pilot study
Controlled Observation
- Observes in dcontrolled envioronment, manipulation of variables (Ainsworth)
- Better controls of extraneous variables.
- Participants usually know being observed, may not behave naturally, lowers validity.
- Informed consent, confidentiality and right to withdraw are issues.
Questionnaires
- Set of questions, collects data from large sample. Given face-to-face, post, phone, internet, left in public places.
- Reaches sample wuickly and cheaply, large amounts of data, time efficient (researcher not needed), reduces investigator effects, easily analysed data and replicable.
- Social desirability-low validity, low response rate, questions ambigous (researcher not there explaining), Closed questions limit respose and open questions difficult to analyse.
- Privacy, confidentiality, informed consent and protection from harm.
What you need to consider when designing a questionnaire:
- Qualitative or quantitative
- Question wording
- Ethical issues
- Number of questions
- Layout of questions
- Sampling
Interviews
- Researcher asks questions face to face, either structured or unstructured:
Structured:
- Data analysis easy, less risk investigator effects, less training for interviewers, can clarify ambiguity.
- Cannot follow up interesting answers, formal situation people may lie, less full answers.
Unstructured:
- Can follow up issues raised by interviewer, can expand on answers (new insights), more informal, interviewer more sensitive with personal information.
- Interviewer effects, social desirability, high level of training for interviewer, time-consuming, expenisve, difficult analysing qualitative data
Privacy, confidentiality, protection from harm, informed consent, right to withdraw are issues.
What you need to consider when designing an interview:
- Structure
- Sampling
- Recording of data
- Analysing qualitative data
- Question wording
- Ethical Issues
Case Studies
- In-depth study of individual/group of people.
- Provides rich data, high ecological validity, suggests new hypotheses for research, investigates topics otherwise unethical/impractical.
- Difficult replication, generalisability and possible research bias.
- Informed consent, invasion of privacy, right to withdraw and confidentiality are issues.
Aims and Hypotheses
Aim: Statement about purpose of investigation, what researcher trying to discover. Hypotheses: Precise estimate of what will happen in the investigation (experimental hypothesis)
- Directional Hypothesis: Predicts direction of difference.
- Non-Directional Hypothesis: Doesn't predict direction of difference (no direction).
Repeated Measures Design
- Same participants in all conditions.
- Holds individual differences constant, fewer participants.
- Order effects, increased chance demand characteristics, cannot use same stimulus in conditions.
Independent Groups Design
- Different participants in all conditions (randomly allocated group/no bias)
- No order effects, reduce demand characteristics, same set of stimulus in all conditions.
- Individual differences, more participants needed.
Matched Pairs Design
- Participants matched by key participant variables.
- No order effects, reduces individual differences, use same stimulus in all conditions.
- Difficult to decide key participant variables, difficult matching exactly, more participants.
Operationalisation
Operationalisation: Defining how a variable can be measured, making study replicable.
Pilot Studies
A small scale study , on a small sample created before main study testing the methodology and to decrease any problems with:
- Sampling method
- Design
- Instructions to participants
- Choice of stimulus etc...
Can be created before any type of study.
Extraneous Variables
Important to try and eliminate as many extraneous variables as you can, can never be all gone.
- Participant variables: intelligence, age, genderm perosnality etc. choosing appropriate design (matched and repeat avoid individual differences, independent avoids bias)
- Demand characteristics: cues helping work out the meaning of experiment altering behaviour. (single blind)
- Experimenter effects: characteristics of investigator that may affect results (double blind)
- Situational variables: temperature, time, lighting, stimulus material (standardisation, conditions same for everyone)
Reliability & Validity
Reliability:
- consitency of results, if repeated you get the same results (high) and vice versa.
Validity:
- Measuring what we want to.
- Internal: If the outcome is the result of the manipulated variables (IV not extraneous variables)
- External: Findings being generalised to outer world (outside research settings)
- Population Validity: Generalised to other people
- Ecological Validity: Generalised to other settings
Ethical Issues
BPS Code Of Ethics:
- Deception: Information not withheld, shouldn't be misled.
- Informed Consent: Aims of research clear, parental consent for children under 16.
- Right to Withdraw: Can withdraw at any time of the study, any data has to be destroyed.
- Protection from Harm: Responsibility protect participants from psychological/physical harm. Not exposed to any higher risk than everyday life.
- Debriefing: Given full explanation after study is completed.
- Confidentiality: Data kept anonymous, if cant participants must be clearly informed before.
- Privacy: Keep paricipants privacy making sure information leading to person is not published
- Colleagues: When suspected colleague follows unethical procedure, raise concerns.
- Giving Advice: Only offer advice within own expertise, otherwise employ someone else.
Guidelines, sometimes bent for meaningful research.
Sampling and Participants
Random:
- everyone in target population has equal chance being picked.
- each member identified then sample technique used.
- cannot guarantee representative sample (most likley however), difficult as you must have access to all members (usually only possible with small target), if anyone drops out sample is less random.
Oppotunity:
- People readily available to researcher.
- Approaches available people willing to do research.
- Not representative(find only certain people that time of day), Feel obliged(dont want to but do), very convenient.
Volunteer:
- Participants self-select.
- Researcher advertises for volunteers and volunteers reply.
- Volunteer bias (not representative), useful when highly specific volunteers needed.
Quantitative Data
Graphs:
- Bar chart: nominal data, vertical bar represents different category (x-axis), frequency (y-axis)
- Histograms: ordinal or interval data, Units of measurement(single or grouped) on x-axis. Frequency represented by ares of vertical bars.
- Freqency Polygon: Alternative to histogram, useful when showing 2 sets of data on 1 graph.
- Scattergram/graph: Relationship between 2 variables (correlations), 1 variable shown on y-axis other on the x-axis. Closer to straight line points are-stronger correlation.
Qualitative Data
- Cannot be expressed in numerical values.
Content Analysis: systematic research technique for analysing qualitative data.
- Decide material sampling.
- Decide the themes and categories that may emerge in these materials.
- Create coding system based on themes/categories.
- Collect large sample of chosen samply materials.
- Coders given sample material, read and categorise items according to coding units (words, themes, character)
High validity as gathered in natural setting.
Interpretation can be subjective, can be inconsistent and take own meaning to coding system so often unreliable.
Improve reliability by having mutiple coders, discussing coding units together. Correlational techniques are used to check reliable coders.
Central tendency and dispersion
Central tendency:
- Mean: add all scores divide by number of scores. (+: sensitive measure, takes all scores into account.-: distorted by anomalie, only used for data at least interval)
- Median: scores in order then take middle value. (+:ordinal or interval data, unaffected by extreme scores.-: not useful in small sets of data, unrepresentative of data if general clutter high or low or a few values.
- Mode: frequent occuring value.(+: easy to calculate, nominal data, unaffected by extreme scores.-: not useful in small sets of data, not useful when more than one modal score in set, doesn't tell us about other scores in set, may not be central)
Dispersion:
- Range: difference between highest and lowest score. (+: quick, easy to calculate. -: easily distorted from anomalie)
- Standard Deviation: average amount score differs from mean (+: account of all scores. -: more complicated than range, only with interval scale measured data)
Comments
No comments have yet been made